Re: SUO: Criteria that an ontology must satisfy
Jim,
I regard the testing to be of the same nature as the testing
of any hypothesis in science. Somebody proposes some theory
of categories for the SUO and says that he or she believes it
to be adequate to meet the requirements.
Then anybody who doubts that can ask the proposer(s) to define
any suggested term. If the proposer(s) cannot come up with
a suitable definition, then they have to correct the theory,
extend it, or modify it in some way. It's very much like the
task of debugging a program to fix bug reports.
>... What does
>it mean for the SUO itself to have this ability? How is this ability
>demonstrated?
By trial and error. It has to provide a sufficient collection
of predicates with their axioms and definitions to be able to
define any term that anyone can throw at it.
>Second, since you suggest "only" the ability to define, and not the
>definitions themselves, your criteria seem to leave open the possibility of
>adding new definitions. Is this a correct reading of what you say?
Absolutely. It has to grow in the same way that any scientific
theory or any widely used computer system grows and develops to
support new applications.
>Third, you mention "The ability to support inferences ... that agree with
>established usage...." How is this ability to be demonstrated? By looking
>at "all possible" inferences that the SUO generates and then determining
>that such inferences do indeed accord with established usage? Is that
>desirable or even possible?
No. The proposer(s) propose the theory, and the skeptic(s)
challenge it. If it successfully withstands all challenges
(perhaps with some occasional time out for repairs), then it
is considered to meet the requirements. But there is always
a possibility that someone might come up with a new challenge.
Then the proposer(s) have to go back to the drawing board to
find some appropriate modification or extension.
>Fourth, what other criteria might be suggested, specific to the SUO
>efforts, that allow the following question to be answered?
>
> How will the SUO group determine that it has successfully met its
>goals?
When all the challengers give up -- and like any science, that
process might be never ending. But if the approach is good,
any stage of development is adequate for a wide range of
applications and later stages get better.
John Sowa